Spatial Characteristics of Lightning Relative to Supercell Kinematics and Microphysics

Spatial characteristics of lightning relative to supercell kinematics and microphysics

Sarah M. Stough, Lawrence D. Carey Christopher J. Schultz, Daniel J. Cecil Department of Atmospheric Science Earth Science Office University of Alabama in Huntsville NASA Marshall Space Flight Center Huntsville, Alabama, United States Huntsville, Alabama, United States [email protected]

Abstract— Rapid increases in total lightning flash rate, volume, updraft velocity, and measures of updraft termed lightning jumps, are often observed prior to high-impact size, such as the volume of the updraft in which convective phenomena. Relationships between lightning jumps and substantial increases in quantifiable bulk updraft properties vertical velocities are greater than or equal to -1 have been shown. However, the details that support these 10 m s [e.g., Carey and Rutledge 1996, 2000; Lang relationships are not well resolved, particularly with respect to and Rutledge 2002; Deierling et al. 2008; Deierling variability of flash rate trends in response to kinematic properties and specific, process-based connections with high- and Petersen 2008; Calhoun et al. 2013, 2014]. impact phenomena. This study addresses detail within the Meanwhile, quantified rapid increases in flash rate, complex relationships between lightning, kinematics, and referred to as lightning jumps, have been shown to microphysics via observations of a supercell thunderstorm. be related to more significant changes in the updraft Correspondence between flash rates and bulk updraft characteristics agreed with established relationships while flash of a thunderstorm [Schultz et al. 2015, 2017]. In properties varied between significant changes in flash rate. particular, quantified lightning jumps correspond Dominant flash size and altitudes of flash initiation during with changes in the 10 m s-1 updraft volume and lightning jumps corresponded spatially with a complex maximum updraft speeds that are roughly four and combination of updraft characteristics and graupel distribution. Flashes associated with a rapid decrease in flash rate showed a five times greater, respectively, than those that occur weaker spatial relationship with the updraft. in association with non-jump increases in flash rate [Schultz et al. 2017]. Keywords—supercell; Doppler radar; thunderstorm microphysics Relationships between flash rates and updraft properties have physical roots in the suggested role of the updraft in the non-inductive charging I. INTRODUCTION mechanism, storm-scale charge separation, and Observations relating rapid increases in total subsequent flash production [Reynolds et al. 1957; lightning flash rate to high-impact phenomena have Takahashi 1978; Carey and Rutledge 1996, 2000; been frequently reported in the literature, motivating Deierling and Petersen 2008]. However, the study of lightning as a metric of thunderstorm complexities within the physical underpinnings of intensity [e.g., Williams et al. 1999; Goodman et al. these relationships have not been well resolved. For 2005; Schultz et al. 2009, 2011, 2015, 2017; Darden instance, the relative roles of kinematics and et al. 2010; Gatlin and Goodman 2010]. Flash rates microphysics in the generation of a lightning jump, have generally been shown to relate to properties connections between lightning and specific such as thunderstorm graupel mass or graupel thunderstorm processes contributing to severe

This work was supported by National Aeronautics and Space Administration (NASA) Severe Storms Research funding (NNH14ZDA001N), provided to 1 University of Alabama in Huntsville authors under contract from the NASA Marshall Space Flight Center (NNM11AA01A) weather, and the mechanisms behind rapid decreases 0157 UTC and 0212 UTC. No other reports of in flash rate, including those occasionally observed severe weather were documented during the analysis prior to tornadogenesis, are not well understood. In period. particular, lightning’s connection with and response In addition to providing a well-sampled case, the to vertical kinematics have not been deeply explored supercell mode represented by this storm presents with respect to downdraft-generated phenomena. the opportunity to evaluate properties of interest Investigation of these complex, interconnected within quasi-steady storm structure. Additionally, properties and processes related to high-impact the supercell storm mode includes prevalent phenomena requires more detailed analyses at finer kinematic regions for evaluation alongside well- spatial scales than reported in many studies of bulk developed microphysical fields and lightning thunderstorm properties. properties. The main updraft, forward flank The most productive path to understanding downdraft (FFD), and rear flank downdraft (RFD) details that connect lightning and thunderstorm are the primary vertical kinematic regions identified processes ultimately requires a combination of well- in supercell structure. While the RFD is thought to observed thunderstorm events and high-resolution initially result from storm flow and pressure analysis of process afforded by numerical pertubations, both the RFD and FFD are supported simulations. This study serves as an introductory by a combination of precipitation loading and latent analysis of a well-observed thunderstorm case in heat exchange as a result of microphysical which the details of flash properties, microphysics, processes [Hookings 1965; Lemon and Doswell and kinematics are considered during significant 1979; Srivastava et al. 1985, 1987; Knupp 1987, trends in lightning flash rates. Information from 1988; Vonnegut 1996; Tong et al. 1998; Naylor et available radar and lightning data is discussed, al. 2012]. where results include detailed descriptions of lightning, microphysics, and kinematic behaviors; their variation in space and time; and possible connections that may be evaluated through additional observations and future implementation of a modeling component.

II. DATA AND METHODS A. 01 April 2016 Supercell Case A supercell that occurred on 01 April 2016 in northern Alabama serves as the thunderstorm of interest for this study. It was observed by a suite of instrumentation during a Verification of the Origins of Rotation in Tornadoes Experiment – Southeast (VORTEX-SE) intensive operations period, including multiple Doppler radars and the North Alabama Lightning Mapping Array (NALMA). Instrumentation platforms within the sampling domain relevant to this study are shown in Fig. 1. While data were collected within the sampling Fig. 1: VORTEX-SE sampling domain and relevant instrumentation. The domain for several hours prior to, during, and ARMOR and MAX radar are plotted as blue and red dots, respectively, while following passage of the supercell, the analysis the dual-Doppler lobes determined from 30° beam crossing are marked by the purple circles. The 10 LMA sensors that usually comprise the NALMA within period of the storm is limited to the 0100 UTC to northern Alabama and southern Tennessee are plotted as green crosses and the 0200 UTC during which time two nearby radars mobile LMA sensors added to the network for the VORTEX-SE project are plotted as blue crosses. The path of the storm of interest is plotted as a red line, were providing observations. The supercell where this analysis focuses upon the period during which the storm propagated produced an EF2 tornado that was reported between through the southern dual-Doppler lobe.

B. Lightning Data and Processing Methods (FID), or the number of lightning flashes that occur The NALMA typically consists of a within a given grid pixel, and flash extent density configuration of twelve sensors located in northern (FED), or the number of lightning branches ot pass Alabama, southern Tennessee, and northwestern through a given grid pixel, were gridded with Georgia that detect very high frequency (VHF) 1.0 km horizontal and 0.5 km vertical grid spacing. emissions during the propagation of lightning For gridded products and subsequent flash-based channels [Rison et al. 1999]. However, six mobile analyses, only flashes with a minimum of 10 sources LMA sensors were added to the standard NALMA were considered to reduce the likelihood of domain during the VORTEX-SE field project, including noise erroneously identified as flashes. providing additional network sensitivity. A total of Processed lightning flash data were associated 16 of the 18 available sensors were active during the with the supercell according to the methods analysis period of this supercell, all located within described in Stough et al. [2017]. Briefly, using the northern Alabama and southern Tennessee [NASA Warning Decision Support System – Integrated 2001]. VHF sources detected by LMA sensors are Information (WDSS-II) software, storm objects located in time and space utilizing a time of arrival were identified using environmental and radar technique [Thomas et al. 2004]. While data from a reflectivity data [Lakshmanan et al. 2007]. Storm minimum of four sensors are required to resolve the boundaries based on these objects were used to time and location of a source, six or more sensors isolate lightning flashes associated with the storm of are typically utilized in practice to reduce the interest. Flashes collected via storm tracking were inclusion of noise. Source data are subject to binned in 2-min intervals and processed to location errors on the order of 1 km in the vertical determine 1-min flash rate and the presence of any and 500 m in the horizontal within 100 km of the lightning jumps and lightning dives, or rapid LMA network center, outside of which they may decreases in flash rate, according to the Schultz et al. grow to surpass convective scales [Koshak et al. [2009, 2015] 2s lightning jump algorithm (LJA). A 2004; McCaul et al. 2005]. The storm of interest for lightning jump (dive) is determined for a 2-min this study remained within 100 km of the NALMA interval during which the change in flash rate is center for the duration of its analysis. greater (less) than at least two standard deviations of NALMA source data were clustered into flashes the change in flash rate observed over the prior with the python-based open-source “lmatools” 10-min period. Additionally, the minimum flash rate software [Bruning 2013], in which a set of spatial must be at least 10.0 flashes per minute (fpm) and a and temporal criteria are employed to group sources jump cannot follow another by fewer than 6 min. In into flashes according to the DBSCAN technique the case of a lightning dive, the flash rate must have [Bruning 2013; Fuchs et al. 2015, 2016]. For this been at least 10.0 fpm within 4 minutes prior to the study, a spatial threshold of 1 km and a temporal dive. Note that lightning dive flash rate requirements threshold of 0.3 s were used to identify sources are not implemented in the traditional Schultz et al. within a flash, consistent with the McCaul et al. [2009, 2015] 2s LJA. The “s-level” associated [2009] methods historically utilized with NALMA with the jump (dive) is the number of standard data [e.g., McCaul et al. 2009; Schultz et al. 2009, deviations by which the increase in flash rate at the 2011]. A maximum flash duration of 3 s was also time of the jump exceeded (fell below) the recent imposed. Sources were additionally required to have history of change in flash rate. In the case of a been detected by a minimum of six sensors in order lightning jump, the s-level reflects the intensity of to be considered for flash clustering. the jump, and by extension, the change in intensity of the updraft as inferred through a property such as As part of flash clustering, the lmatools software -1 maximum speed or 10 m s volume [Schultz et al. is also capable of providing a number of flash 2017]. properties, including the time and three-dimensional locations of flash initiation, the number of sources C. Radar Data and Processing Methods within each flash, flash area and volume. Lmatools The supercell of interest was best sampled by the optionally provides post-processed gridded products fixed-site C-band Advanced Radar for of flash-clustered data. Flash initiation density Meteorological and Operational Research

(ARMOR; Schultz et al. 2012; Knupp et al. 2014), considered include differential reflectivity (ZDR), though was also within adequate range of the specific differential phase (KDP), and copolar cross Mobile Alabama X-band radar to allow dual- correlation coefficient (rHV) in addition to the Doppler analysis for vertical wind retrievals. Data horizontal reflectivity (ZH) and temperature data collected from the ARMOR are primarily discussed inputs. Prior to HID analysis, ARMOR data were herein, while Doppler velocity data collected from corrected for attenuation and processed to obtain the MAX radar were also utilized. specific differential phase (KDP) through in-house Prior to conducting kinematic analysis, the implementation of the Bringi et al. [2001] methods. ARMOR and MAX velocity data were manually The Dolan et al. [2013] HID algorithm is included edited and dealiased using the National Center of as part of the CSU Radar Tools python software Atmospheric Research (NCAR) SOLOIII software package [Dolan et al. 2002]. To fascilitate their use [Oye et al. 1995; Vacek 2017]. These data were with the python-based implementation of the then gridded to a common Cartesian coordinate algorithm, ARMOR data were imported and gridded system with a 1 km horizontal and 0.5 km vertical with 1.0 km horizontal and 0.5 km vertical grid grid spacing using the NCAR Radx software spacing utilizing the Python Advanced Radar Tools [Heistermann et al. 2014; Vacek 2017]. Dual- (Py-ART) package prior to HID analysis [Helmus et Doppler analysis was then performed using the al. 2016]. In addition to qualitative analysis, gridded NCAR Custom Editing and Display of Reduced HID data were used to compute graupel volumes Information in Cartesian space (CEDRIC) software within the storm, derived from the number of grid for the retrieval of vertical velocity data [Mohr et al. pixels corresponding to either graupel category. 1986; Miller and Frederick 1998; Vacek 2017].

Microphysical analysis was accomplished III. RESULTS through use of hydrometeor identification (HID) via the Dolan et al. [2013] HID algorithm with ARMOR During the analysis period from 0100 UTC to data. The HID algorithm identifies the dominant 0200 UTC, supercell lightning flash rates were precipitation-sized hydrometeor type in a radar gate observed between 1.5 fpm and 59.0 fpm, a time or grid pixel based on weighted polarimetric radar series of which is shown in Fig. 2. Three lightning data and temperature data inputs that are evaluated jumps occurred during this period at 0130 UTC, using a fuzzy logic scheme. There are ten possible 0136 UTC, and 0142 UTC, as well as one lightning hydrometeor types, including melting hail/large dive at 0146 UTC. While these significant lightning raindrops, hail, high-density graupel, low-density flash rate trends constitute the focus of analysis, graupel, aggregates, vertical ice, ice crystals, wet traditional bulk properties of the thunderstorm are snow, drizzle, and rain. The polarimetric data first addressed.

Fig. 2. Time series of flash rate and LJA information. Flash rate (black histogram) and s-level (orange line) are plotted against the left and right axes, respectively. Times at which a lightning jump occurred are marked in red and times at which a lightning dive occurred are marked in blue.

-1 A. Bulk Kinematic Properties 10 m s updraft volume, a second lightning jump Vertical wind data were evaluated during the occurred at 0136 UTC. Updraft volume analysis periods at time intervals corresponding to characteristics are not available after 0136 UTC the ARMOR volume scan start times. Note that the given the poor data quality. However, these observed trends between flash rate and updraft speed quality of dual-Doppler analysis became markedly -1 degraded beginning at 0136 UTC as the MAX radar and 10 m s updraft volume agree well with those signal was attenuated by heavy precipitation. After quantified in previous work [e.g., Schultz et al. 0136 UTC, relative extrema and quality of vertical 2017]. motion may be assessed, though data are not B. Spatial Behavior of Lightning and Microphysics suitable for quantitative analysis. with Respect to the Updraft To allow comparison of broad updraft properties The correspondence of temporal trends in flash with flash rates as accomplished in other studies, rate, graupel volume, and vertical velocity have been maximum updraft velocity and 10 m s-1 mixed- discussed extensively in the lightning jump phase updraft volume were identified at each literature, though their spatial relationships are not gridded plane above ground level (AGL), illustrated frequently reported. To consider how graupel fields in Fig. 3. For reference, the mixed-phase region is and lightning flashes relate in space to the updraft, defined between the heights of 0°C and -40°C, as well as to each other, the locations of these correponding to 3.9 km and 9.9 km for this storm properties were evaluated through time with respect environment. Additionally, the height of -10°C was to the three-dimensional location of the maximum located at 5.2 km. vertical velocity of the updraft region. The relative position and intensity of the updraft For comparison of graupel regions with the is of interest when considering the response of location of the updraft, the median x-coordinate lightning flash properties to kinematics. Generally, position and y-coordinate position were identified at the 10 m s-1 updraft volume remained at or below 36 each altitude at each radar analysis time. The median km3 until 0126 UTC, nearest to the time of the first location point was evaluated against the location of lightning jump. At 0131 UTC, total 10 m s-1 updraft the maximum updraft to determine a direction with volume relaxed from a maximum of 102 km3, respect to the updraft as well as the range of the mostly distributed between 5.5 km and 9.0 km, to a graupel field from the maximum updraft. These volume of 57 km3, mostly distributed between results are shown in Fig. 4. Additionally, the 4.5 km and 6.0 km. Despite the general decline in positions of large values of FID and FED were magnitude, vertical extent, and altitude of the considered. To determine regions of maximum FID

Fig. 3. Time-height plots of maximum vertical velocity within the main updraft region of the supercell (color-filled, left) and of 10 m s-1 updraft volume (color- filled, right). Times of poor dual-Doppler quality are shaded in gray. Within each panel, altitudes of 0°C, -10°C, and -40°C (horizontal dashed blue lines) and the altitude of the maximum vertical velocity within the updraft column at each time (dark red line, outlined in white) are also marked. Lightning jump (red) and lightning dive (blue) times are shown as vertical lines. Additionally, the total updraft volume is plotted in violet within the right panel along the right vertical axis.

Fig. 4. Time series of the position (top) and range (middle) of the median graupel field location with respect to the maximum updraft, along with graupel volume within the storm (bottom) with height. The altitude of the maximum vertical velocity within the updraft column at each time (dark red line) is also included in the middle panel. Heights of 0°C, -10°C, and -40°C (horizontal dashed blue lines) and markers at the times of lightning jumps and dives (red and blue vertical lines) are shown in each panel. and FED, the 90th percentile value of each property maximum updraft from 4.5 km to 7.5 km. At this was calculated at each altitude level at each time. time, graupel distance from the maximum updraft The median x-coordinate position and y-coordinate position decreased substantially from over 7.5 km th position of values greater than or equal to the 90 to less than 3.0 km through the mixed-phase region, percentile values were evaluated against the location with shortest distances coinciding with the height of of the maximum updraft, as was done with the the maximum updraft. Simultaneously, the ranges median graupel location. Results from analysis of of large FID and FED values from the updraft the position of lightning properties are shown in decreased to less than 5.0 km and were comparable Fig. 5. with graupel ranges from the updraft in the mixed- While graupel volume increased steadily within phase region. In contrast, the nearest large FID and the lower mixed-phase region of the storm, the first FED values were slightly displaced above the lightning jump at 0130 UTC occurred during an altitude of the maximum updraft by approximately increase in graupel volume near the -10°C level and 2.0 km and more so above the altitudes of the coincident with an increase in height in the greatest updraft volumes at this time (Fig. 3),

Fig. 5. Time series of the position (top) and range (middle) of the median significant flash initiation density (FID, left) and flash extent density (FED, right) locations with respect to the maximum updraft. Here, significant FID and FED were considered to be values greater than or equal to the 90th percentile value observed at each altitude and time (bottom). The altitude of the maximum vertical velocity within the updraft column at each time (dark red line) is included in the middle panels. Heights of 0°C, -10°C, and -40°C (horizontal dashed blue lines) and markers at the times of lightning jumps and dives (red and blue vertical lines) are shown in each panel. consistent with observations made by Schultz et al. 90th percentile values of FID and FED were [2015, 2017]. With respect to relative position, it observed between 7.5 km and 9.5 km. appears that graupel became more oriented to the Spatial characteristics of lightning and graupel west/southwest of the updraft, though this properties with respect to the updraft location at the observation is likely an effect of the proximity of time of the third lightning jump showed some the graupel to the updraft versus a true change in similarities as well as differences compared with relative directional location. Similar displacements behavior observed during the previous two jumps. were observed in FID in the regions of minimum For instance, the relative distance between the range from the updraft, though to a lesser extent. graupel field and the updraft continued to increase, Directional displacements and changes thereof were while mixed-phase graupel volume remained not as evident in FED. consistent. Maximum FID and FED values At the time of the second lightning jump at displayed a second period of relative maxima near 0136 UTC, the relatively small distances between the 7.5 km level, while their locations became the maximum updraft and graupel and lightning closer to the location of the maximum updraft than property fields had expanded from 1 km to 3 km in observed during the previous jump at ranges of less the mixed phase region to values of approximately than 7 km. 5 km to 7 km. Their relative positions also shifted to During the single lightning dive that occurred at the north/northeast of the updraft. However, graupel 0146 UTC, the distance of graupel from the main volume increased in magnitude by approximately 3 updraft decreased and transitioned further to the 50 km per level and expanded with altitude within east of the updraft. While a similar displacement the mixed-phase region at this time. Maxima in the was not discernible in the lightning FID and FED location data, the distance between substantial FID

and FED and the main updraft increased to jump periods (length < 15 km), with the majority of approximately 15 km throughout the mixed-phase small flashes occurring between 7 km and 9 km. levels. Consistent with a rapid decrease in lightning, The relative distributions of flash sizes greater than 90th percentile FID values also decreased. However, 10.0 km were similar for periods from 0120 UTC increased 90th percentile FED values observed at through 0140 UTC, including at the times of the time of the third lightning jump persisted and lightning jumps within that range. However, at extended through more of the lower mixed-phase 0142 UTC when the third jump occurred, the region, though were located in decreasing proximity relative number of flashes ≥ 10 km decreased by to the east/northeast of the updraft. approximately half of the percentage observed in earlier periods. The number of flashes during this C. Flash Properties During Periods of Substantial period increased, consistent with the observation of Flash Rate Trends a lightning jump. It is worth noting that while most Consideration of the spatial components of of these were small flashes likely near to the updraft lightning flashes with respect to kinematics implied as suggested in Fig. 5, there were also a greater a degree of variability in flash properties between number of larger flashes observed between 6.0 km distinct lightning jumps. Flash initiation locations and 8.0 km. During the period of the lightning dive and flash sizes were more closely examined to at 0146 UTC, while the majority of flashes were further explore the characteristics of their variation. still ≤ 6 km in length, relatively more of these Flash size and altitude of initiation of flashes flashes were clustered at lower altitudes between within 2-min periods were examined for 10 min 4.0 km and 6.0 km as opposed to the 7.0 km to prior to the first lightning jump through 2 min after 9.0 km range observed during the jumps. This the lightning dive (Fig. 6). Though a small number vertical distribution persisted through 0148 UTC. of larger flashes (length ≥ 20 km) occurred near Generally, between the time of the third jump at the -10°C altitude of the mixed-phase region 0142 UTC and through the period after the dive at sporadically during the analysis period, most flash 0148 UTC, the number of flashes with lengths > 10 lengths were typically smaller during lightning km also increased throughout the mixed-phase

Fig. 6. Distributions of flash length within 2-min periods (top) and distributions of flash length with altitude within 2-min periods (bottom). The relative percentage of flashes with a length greater than 10 km are annotated in the top rows of each panel. The number of flashes within each 2-min bin are annotated in the bottom rows of each panel. The number of flashes of a given length that occurred at a given altitude are colored according to scales on the right, per row. Note that the scales corresponding to each row are different. The 2-min intervals corresponding to the times of the jumps (dive) are denoted by red (blue) text.

8 region. The proximity of lightning initiation points to the secondary updraft as well as their spatial dispersion It has been suggested that flash size and location relative to the weaker vertical motion indicate that respond to the coupled nature of the distribution of lightning was responding to the new updraft charge regions relative to a) microphysical location and associated microphysics. After constituents and b) kinematic texture associated 0136 UTC, quantitative kinematic information is with strong vertical motion [Bruning et al. 2007; not available. However, similar observations from Bruning and MacGorman 2013; Schultz et al. additional data may provide insight into the utility 2015]. Evaluation of lightning properties in the of the spatial information of lightning to inform context of these convective components may about updraft growth processes as a complement to facilitate understanding of relative roles in the radar interpretation. nature of flash production with respect to flash rate trends. The third lightning jump occurred at 0142 UTC when quantitative updraft information was not Combining the spatial information of the available, though updraft location could still be hydrometeor fields and flash properties at the time loosely ascertained. Fig. 5 indicates that most of the first jump at 0130 UTC, Figs. 7a and 7b show lightning at the time of this jump was associated numerous larger flashes at 5.5 km to 6.0 km at the with the main updraft, where Fig. 7e and Fig. 7f periphery of the graupel region to the north of the suggest that this was the newly formed updraft seen updraft. Meanwhile, the majority of smaller flashes in Fig. 7d. At the time of the third jump, a large at 7.5 km to 8.0 km occurred in close proximity to -1 number of small flashes were observed near the the 10 m s updraft volume, also along the position of the updraft, in close proximity with the periphery of the low-density graupel observed at low density graupel region observed in Fig. 7f. As 7.5 km. These depictions support information from suggested by the information in Fig. 6, a large Figs. 4 and 5 indicating that greater flash initiations number of larger flashes were also observed along a were likely collocated with the primary graupel line from NNW of the main updraft to E of the main regions as evidenced by their similar close updraft. These larger flashes occur between 5.5 km proximity to the updraft maximum location. and 7.5 km, and appear to be associated with a At the time of the second jump at 0136 UTC, relatively expansive region of graupel near 6.0 km. Figs. 4 and 5 indicated that graupel and flash Flash properties of the lightning dive at regions had shifted to the ENE of the maximum 0146 UTC and flash data from the subsequent updraft and had also increased somewhat in range 2-min period at 0148 UTC are shown in Figs. 8a from the updraft. Figs. 7c and 7d show that the and 8b. During this period, a relatively large graupel region had expanded, particularly within the number of small flashes was observed distributed 5.0 km to 6.0 km levels of the mixed-phase region, throughout the storm, as indicated by the distance while most lightning flashes occurred either along from the updraft seen in Fig. 5 and the relative the northeastern periphery of the graupel near distribution of flash size observed in Fig. 6. 5.0 km or further aloft near 8.0 km to 9.0 km. While Generally, flashes appear to have responded less to the maximum updraft was located to the southwest the updraft, though in absence of quantitative of the storm, a new, secondary updraft region had kinematic information, it is not possible to ascertain begun to develop to the northeast between 7.5 km any measure of the strength of the updraft at this and 10 km. The weaker secondary updraft was time. However, qualitative flash properties indicate evident at the 8.5 km level with a small region of -1 that the updraft was not supporting or influencing 10 m s updraft volume (Fig. 7d). Though more flash behavior during this period as observed during dispersed relative to the new updraft than flashes the lightning jumps. It is suggested that the observed near the main updraft during the first lightning dive observed in this instance was a lightning jump, flash initiation locations were closer response to the decreased capability of the updraft to the new secondary updraft than the main updraft. to support mixed-phase updraft growth and particle-

Fig. 7. Plots of lightning flash initiation locations in two minute intervals corresponding to the lightning jumps at 0130 UTC (a,b), 0136 UTC (c,d), and 0142 UTC (e,f). Flash points are marked as circles, colored by altitude of initiation and sized according to relative length. Flashes are plotted over planar images of HID at corresponding ARMOR volume times at various altitudes, marked above the left-most color bar in each panel. The 10 m s-1 updraft contour is plotted (red line) along with the location of the main updraft (dark blue cross). When a secondary updraft is present (c,d), its location marked (light blue cross) in addition to that of the main updraft. scale charge separation through its influence on altitudes during and following the time of the rebounding collisions. This behavior was coincident lightning dive. Simultaneously, graupel volume with mesocyclogenesis inferred from Doppler trends indicate that graupel may have been velocity data, where the reported tornado was descending within the storm, particularly as values associated with the new mesocyclone [Vacek 2017]. between 4.0 km and 6.5 km decreased while values Further, flashes appear to have transitioned to lower between 3.5 km and 4.0 km increased.

consistent with observed spatial changes in FED in

Fig. 8: As described in Fig. 7 for the lightning dive period at 0146 UTC (a) and the subsequent 2-min period at 0148 UTC (b).

IV. SUMMARY AND DIRECTION OF CONTINUING particular. WORK Placing these properties in the context of This study considered details within bulk trends microphysics, it was observed that dominant sizes in lightning, kinematic, and microphysical and preferential altitudes of flash initiation were thunderstorm properties observed in a supercell associated with kinematics as well as the spatial thunderstorm as a preliminary evaluation of the distribution of graupel regions. Moreover, dominant complexities between lightning and thunderstorm flash sizes and apparent preferential altitudes of microphysics and kinematic relationships. flash initiation during lightning jumps appeared to Prior to addressing flash properties and spatial respond to a complex combination of the presence relationships with microphysical and kinematic of graupel near regions of vertical velocity. As a fields, it was first noted that broad trends between specific example, lightning that occurred during the lightning flash rates and updraft properties were second observed lightning jump at 0136 UTC was consistent with those documented in the literature. spatially associated with a region of developing Upon examining the spatial nature of the evolution vertical velocity aloft and nearby graupel rather than of lightning properties and microphysics with with graupel fields in the vicinity of the stronger, respect to the updraft, variability in properties of broader updraft. lightning flashes associated with lightning jumps Lightning dives have been observed for some emerged. Specifically, it was observed that while time, particularly with respect to tornadogenesis regions in which many flashes formed and [e.g., Steiger et al. 2007], though their physical propagated corresponded closely in space with nature has only recently been addressed [Vacek mixed-phase graupel regions, their locations each 2017]. Characterization of flash properties tended to vary with respect to the location of the associated with lightning dives is therefore limited. main updraft at the times of different lightning However, these analyses indicate that a lightning jumps. Examing flash properties during each period dive is not only characterized differently from in more detail, variations in flash size with altitude lightning jumps in terms of the numbers of flashes at the times of different jumps were apparent, that occur, but may also differ with respect to how flashes are distributed within the storm. Specifically, 11 while the distribution of flash size was similar during the dive as observed at other periods during Carey, L. D., and S. A. Rutledge, (2000), The relationship between the storm, smaller flashes were not clustered near precipitation and lightning in tropical island convection: A C-band the main updraft but rather were more distributed polarimetric radar study. Mon. Weather Rev., 128, 2687–2710, throughout graupel regions that extended to lower doi:10.1175/1520-0493(2000)128<2687:TRBPAL>2.0.CO;2. altitudes. Darden, C. B., D. J. Nadler, B. C. Carcione, R. J. Blakeslee, G. T. Stano, and D. E. Buechler, (2010), Utilizing total lightning information to Additional analyses are necessary as part of diagnose convective trends. Bull. Am. Meteorol. Soc., 91, 167–175, future work to characterize the variability of doi:10.1175/2009BAMS2808.1. lightning during significant trends in flash behavior. Goals of ongoing work are to clarify the respective Deierling, W., and W. A. Petersen, (2008), Total lightning activity as an indicator of updraft characteristics. J. Geophys. Res., 113, D16210, roles of microphysics and kinematics in the doi:10.1029/2007JD009598. generation of a lightning jump, thereby refining Deierling, W., W. A. Petersen, J. Latham, S. Ellis, and H. J. Christian, (2008), understanding of how these lightning trends relate to The relationship between lightning activity and ice fluxes in thunderstom intensity. Additionally, better thunderstorms. J. Geophys. Res., 113, D15210, doi:10.1029/ understanding of the physical basis of lightning 2007JD009700. dives requires further observation to understand the Dolan, B., B. Fuchs, K. Wiens, R. Cifelli, L. Carey, T. Lang and Coauthors, information they may provide concerning (2002), CSU Radar Tools, accessed 2016. [Available online at https://github.com/CSURadarMet/CSU_RadarTools.] thunderstorm processes. Dolan, B., S. A. Rutledge, S. Lim, V. Chandrasekar, and M. Thurai, (2013), A robust C-band hydrometeor identification algorithm and application to a long-term polarimetric radar dataset. J. Appl. Meteorol. Climatol., ACKNOWLEDGMENT 52, 2162–2186, doi:10.1175/JAMC-D-12-0275.1. We gratefully acknowledge Austin Vacek for Fuchs, B. R., E. C. Bruning, S. A. Rutledge, L. D. Carey, P. R. Krehbiel, and providing dual-Doppler analysis data as an essential W. Rison, (2016), Climatological analyses of LMA data with an open- component of this research. source lightning flash-clustering algorithm. J. Geophys. Res. Atmos., 121, 8625–8648, doi:10.1002/2015JD024663.Received.

Fuchs, B. R., and Coauthors, (2015), Environmental controls on storm REFERENCES intensity and charge structure in multiple regions of the continental United States. J. Geophys. Res. Atmos., 120, 6575–6596, doi:10.1002/2015JD023271. Bruning, E. C. (2013), Streamed clustering of lightning mapping data in Python using sklearn, presented at 12th Python in Science Conference Gatlin, P. N., and S. J. Goodman, (2010): A total lightning trending algorithm (SciPy), vol. 2, Austin, Texas, United States. to identify severe thunderstorms. J. Atmos. Ocean. Technol., 27 (1), 3– 22, doi:10.1175/2009JTECHA1286.1. Bruning, E. C., and D. R. MacGorman, (2013), Theory and observations of controls on lightning flash size spectra. J. Atmos. Sci., 70, 4012–4029, Goodman, S. J., and Coauthors, (2005), The North Alabama Lightning doi:10.1175/JAS-D-12-0289.1. Mapping Array: Recent severe storm observations and future http://journals.ametsoc.org/doi/abs/10.1175/JAS-D-12-0289.1. prospects. Atmos. Res., 76, 423-437.

Heistermann, M., and Coauthors, (2014), The emergence of open source Bruning, E. C., W. D. Rust, T. J. Schuur, D. R. MacGorman, P. R. Krehbiel, software for the weather radar community. Bull. Amer. Meteor. and W. Rison, (2007), Electrical and polarimetric radar observations Soc., 96, 117–128, doi:https://doi.org/10.1175/BAMS-D-13-00240.1. of a multicell storm in TELEX. Mon. Weather Rev., 135, 2525–2544, doi:10.1175/MWR3421.1. Helmus, J. J. and Collis, S. M., (2016), The Python ARM Radar Toolkit (Py- ART), a library for working with weather radar data in the Python Calhoun, K. M., D. R. MacGorman, C. L. Ziegler, and M. I. Biggerstaff, programming language. J. of Open Research Software. 4 (1), p.e25. (2013), Evolution of lightning activity and storm charge relative to doi:http://doi.org/10.5334/jors.119. dual-Doppler analysis of a high-precipitation supercell storm. Mon. Weather Rev., 141, 2199–2223, doi:10.1175/MWR-D-12-00258.1. Hookings, G. A., (1965), Precipitation-maintained downdrafts. J. Appl. Meteorol., 4 , 190–195. Calhoun, K. M., E. R. Mansell, D. R. MacGorman, and D. C. Dowell, (2014), Numerical simulations of lightning and storm charge of the 29-30 May Knupp, K. R., (1987), Downdrafts within High Plains cumulonimbi. Part I: 2004 Geary, Oklahoma supercell thunderstorm using EnKF mobile General kinematic structure. J. Atmos. Sci., 44, 987–1008, radar data assimilation. Mon. Weather Rev., 142, 140709125303002, doi:10.1175/15200469(1988)045<3965:DWHPCP>2.0.CO;2. doi:10.1175/MWR-D-13-00403.1. Knupp, K. R., (1988), Downdrafts within High Plains cumulonimbi. Part II: Carey, L. D., and S. A. Rutledge, (1996), A multiparameter radar case study Dynamics and thermodynamics. J. Atmos. Sci., 45, 3965–3982, of the microphysical and kinematic evolution of a lightning producing doi:10.1175/1520-0469(1988)045<3965:DWHPCP>2.0.CO;2. storm. Meteorol. Atmos. Phys., 59, 33–64, doi:10.1007/BF01032000. 12

Knupp, K. R., and Coauthors, (2014), Meteorological Overview of the Schultz, C. J., W. A. Petersen, and L. D. Carey, (2009), Preliminary Devastating 27 April 2011 Tornado Outbreak. Bull. Amer. Meteor. development and evaluation of lightning jump algorithms for the real- Soc., 95, 1041–1062, doi: 10.1175/BAMS-D-11-00229.1. time detection of severe weather. J. Appl. Meteor. Climatol., 48, 2543–2563, doi:10.1175/2009JAMC2237.1. Koshak, W. J., and Coauthors, (2004), North Alabama Lightning Map- ping Array (LMA): VHF source retrieval algorithm and error analyses. J. Atmos. Ocean. Technol., 21 (4), 543–558, doi:10.1175/1520- Schultz, C. J., W. A. Petersen, and L. D. Carey, (2011), Lightning and severe 0426(2004)021⟨ 0543:NALMAL⟩ 2.0.CO;2. weather: A comparison between total and cloud-to-ground lightning trends. Wea. Forecasting, 26, 744–755, doi:10.1175/WAF-D-10- 05026.1. Lakshmanan, V., T. Smith, G. Stumpf, and K. Hondl, (2007), The Warning

Decision Support System–Integrated Information. Weather Forecast., Schultz, C. J. and Coauthors, (2012), Dual polarization tornadic debris 22, 596–612, doi:10.1175/WAF1009.1. signatures Part I: Examples and utility in an operational setting. Electronic J. Operational Meteor., 13, 120-137. Lang, T. J., and S. A. Rutledge, (2002), Relationships between convective storm kinematics, precipitation, and lightning. Mon. Wea. Rev., Schultz, C. J., W. A. Petersen, and L. D. Carey, and R. J. Blakeslee, (2015), 130 (10), 2492–2506, doi:10.1175/15200493(2002)130⟨ Insight into the kinematic and microphysical processes that control 2492:RBCSKP⟩ 2.0.CO;2. lightning jumps. Wea. Forecasting, 30, 1591–1621, doi:10.1175/WAF-D-14-00147.1. Lemon, L. R., and C. A. Doswell, (1979), Severe Thunderstorm Evolution and Mesocyclone Structure as Related to Tornadogenesis. Mon. Wea. Schultz, C. J., W. A. Petersen, and L. D. Carey, and R. J. Blakeslee, (2017), Rev., 107 (9), 1184–1197.. Kinematic and microphysical significance of lightning jumps versus nonjump increases in total flash rate. Wea. Forecasting, 32, 275–288, McCaul, E. W., J. Bailey, J. Hall, S. J. Goodman, R. Blakeslee, and D. E. doi:10.1175/WAF-D-15-0175.1. Buechler, (2005), A flash clustering algorithm for North Alabama Lightning Mapping Array data. Conf. on Meteoro- logical Srivastava, R. C., (1985), A simple model of evaporatively driven downdraft: Applications of Lightning Data, San Diego, CA, Amer. Meteor. Soc., Application to microburst downdraft. J. Atmos. Sci., 42, 1004–1023, 5.3, doi:10.1175/1520-0469(1985)042<1004:ASMOED>2.0.CO;2. http://ams.confex.com/ams/Annual2005/techprogram/paper_84373.ht m. Srivastava, R. C., (1987), A model of intense downdrafts driven by the McCaul, E. W., S. J. Goodman, K. M. LaCasse, and D. J. Cecil, (2009), Fore- melting and evaporation of precipitation. J. Atmos. Sci., 44, 1752– casting lightning threat using cloud-resolving model simulations. 1774, doi:10.1175/1520-0469(1987)044<1752:AMOIDD>2.0.CO;2. Wea. Forecasting, 24, 709–729, https://doi.org/10.1175/ 2008WAF2222152.1. Stough, S. M., L. D. Carey, C. J. Schultz, and P. M. Bitzer, (2017),

Investigating the Relationship between Lightning and Mesocyclonic Miller, L. J., and S. M. Fredrick, (1998), CEDRIC: Custom Editing and Rotation in Supercell Thunderstorms. Weather Forecast., 32, 2237– Display of Reduced Information in Cartesian Space. Manual, 2259, doi:10.1175/WAF-D-17-0025.1. Mesoscale and Microscale Meteorology Division, National Center for Atmospheric Research, 130 pp. Takahashi, T., (1978), Riming electrification as a charge generation Mohr, C. G., L. J. Miller, R. L. Vaughn, and H. W. Frank (1986), On the mechanism in thunderstorms. J. Atmos. Sci., 35, 1536–1548, merger of mesoscale datasets into a common Cartesian format for doi:10.1175/1520-0469(1978)035<1536:REAACG>2.0.CO;2. efficient and systematic analysis, J. Atmos. Oceanic. Technol., 3(1), 143–161. Thomas, R. J., P. R. Krehbiel, W. Rison, S. J. Hunyady, W. P. Winn, T.

Hamlin, and J. Harlin, (2004), Accuracy of the lightning mapping Oye, D., and M. Case, (1995), REORDER: A program for gridding radar data array. J. Geophys. Res. D Atmos., 109, 1–34, installation and use manual for the Unix version. National Center for doi:10.1029/2004JD004549. Atmospheric Research, 44 pp. [Available online at http://www.eol.ucar.edu/content/reorder-reference-manual.] Tong, H., V. Chandrasekar, K. R. Knupp, and J. Stalker, (1998), Multiparameter radar observations of time evolution of convective Naylor, J., M. a. Askelson, and M. S. Gilmore, (2012), Influence of low-level storms: Evaluation of water budgets and latent heating rates. J. Atmos. thermodynamic structure on the downdraft properties of simulated Ocean. Technol., 15, 1097–1109, doi:10.1175/1520- supercells. Mon. Weather Rev., 140, 2575–2589, doi:10.1175/MWR- 0426(1998)015<1097:MROOTE>2.0.CO;2. D-11-00200.1.

Vacek, A. (2017), Identifying physical processes in supercells for nowcasting NASA, (2001), North Alabama Lightning Mapping Array data. NASA Global severe weather potential, M.S. thesis, Univ. of Ala. in Huntsville, Hydrology Resource Center, Huntsville, AL, accessed 2016, Huntsville. https://lightning.nsstc.nasa.gov/lma/nalma/data/.

Reynolds, S. E., M. Brook, and M. F. Gourley (1957), Thunderstorm charge Vonnegut, B., 1996: Importance of evaporative cooling in the formation of separation, J. Meteor., 14, 426–436. thundercloud downdrafts. J. Appl. Meteorol., 35, 1378.

Rison, W., R. J. Thomas, P. R. Krehbiel, T. Hamlin, and J. Harlin, (1999), A Williams, E., and Coauthors, (1999), The behavior of total lightning activity GPS-based three-dimensional lightning mapping system: Initial in severe Florida thunderstorms. Atmos. Res., 51, 245–265, observations in central New Mexico. Geophys. Res. Lett., 26, 3573– doi:10.1016/S0169-8095(99)00011-3. 3576, https://doi.org/10.1029/1999GL010856.